CN102663760B - Location and segmentation method for windshield area of vehicle in images - Google Patents
Location and segmentation method for windshield area of vehicle in images Download PDFInfo
- Publication number
- CN102663760B CN102663760B CN201210120685.4A CN201210120685A CN102663760B CN 102663760 B CN102663760 B CN 102663760B CN 201210120685 A CN201210120685 A CN 201210120685A CN 102663760 B CN102663760 B CN 102663760B
- Authority
- CN
- China
- Prior art keywords
- image
- wind
- edge
- car plate
- blocking glass
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Expired - Fee Related
Links
Abstract
The invention discloses a location and segmentation method for a windshield area of a vehicle in images. The location and segmentation method is achieved on the basis of a license plate location technology and image scanning statistics. The location and segmentation method includes the following steps of obtaining images and performing preprocessing; detecting edges of a license plate and binarizing the images; locating the license plate; obtaining an edge coordinate of the license plate, extracting longitudinal coordinate information, and calculating and obtaining length of the license plate in the horizontal direction and width of the license plate in the vertical direction; obtaining information of a lower edge and an upper edge of a windshield; determining an image scanning area for horizontal locating scanning; scanning the images and summing up white pixels of each line; locating a horizontal edge and the other horizontal edge of the windshield; and obtaining an edge coordinate of the windshield and locating and segmenting the windshield area according to the horizontal edge data, the lower edge of the windshield and the upper edge of the windshield. According to the location and segmentation method for the windshield area of the vehicle in the images, accurate location of the windshield area of the vehicle is achieved, and the accuracy and the execution efficiency are high.
Description
Technical field
The present invention relates to a kind of image processing method, be specifically related to a kind of vehicle wind-blocking glass region in image be positioned to the method for cutting apart.
Background technology
Sustainable development along with economic society, multidisciplinary multi-field fusion development becomes the new trend of technical development, the also development of the intelligent direction from the intellectuality of single Transportation Model to the multiple Transportation Model cooperation of comprehensive traffic of the development trend of traffic system.Intelligent transportation system (Intelligence Traffic System, ITS) be exactly to grow up under this background, it has converted conventional traffic system to multimedia processing, the presentation mode that integrates Digital Image Processing, robotization, pattern-recognition to the processing of data.The nineties in 20th century, technical maturations of image capturing system such as traffic video camera and in the widespread use of field of traffic, in the new era of having started traffic intelligence, reach traffic control computer and become a reality thereby make to process transport information by computing machine.In recent years, the development of digital image processing techniques more makes intelligent transportation system be developed rapidly.
In current intelligent transportation system, the technology such as car plate fixation and recognition, vehicle identification are all ripe day by day, during the emerging technologies such as car sign is other, the identification of car color are also in and develop rapidly.This is wherein especially ripe with car plate location technology, and the application of multiple mathematical method makes adaptivity, the accuracy rate of current many Position System of automobile license plate locations all approach perfect.But, when traffic control personnel need to investigate vehicle interior scene, for example, trace traffic hazard troublemaker or check when whether passenger inside the vehicle fastens the safety belt, toward contact, need to confirm by the method for manual observation image.
Therefore, if can image be processed, by wind-blocking glass region, dividing vehicle front, location, highlight interested car inner region, can further reach the objects such as recognition of face in car, the interior scene inspection of car.But, because the not obvious of vehicle wind-blocking glass edges of regions and irregular many mathematical methods and the formula of making in shape are all difficult to better be applied.When adopting region mode, mathematical formalism mode to process, whether the accuracy of a lot of intermediate result parameters in processing procedure still depends on wind-blocking glass edge obvious, and whether image background environment is complicated etc.This just makes the error at direct-detection wind-blocking glass edge larger, and treatment effect is not good.
Therefore improve or design a kind of applicable intelligent transportation system, and make positioning result sharpness of border, processing speed is fast, and the method that the vehicle wind-blocking glass zone location of more realistic demand is cut apart is one and possesses very much challenging work.
Summary of the invention
Goal of the invention of the present invention is to provide and a kind of vehicle wind-blocking glass region in image is positioned to the method for cutting apart, to overcome the drawbacks such as the position success rate that direct detecting method of the prior art exists is low, calculated amount is large, make segmentation result not only have higher success ratio, and the vehicle of more fitting on edge tissues, realistic application demand.
To achieve the above object of the invention, the technical solution used in the present invention is: a kind of vehicle wind-blocking glass region in image is positioned to the method for cutting apart, based on car plate location technology and image scanning statistics, realize, comprise the following steps:
(1) obtain image and carry out pretreatment operation, described pretreatment operation comprises carries out gray processing to image;
(2) with Sobel operator, detect car plate edge, binary image;
(3) with pattern matrix coupling car plate matrix of edge, positioning licence plate;
(4) obtain car plate edge coordinate, extract along slope coordinate information, and calculate obtain car plate in level side the length on each and the width in vertical direction;
(5) line segment of car plate horizontal direction is respectively expanded to a length to the left and right both direction of level, two-end-point place vertical line is wind-blocking glass lower edge border, the line segment of car plate horizontal direction is respectively expanded to 2/3 length to the left and right both direction of level, and two-end-point place vertical line is wind-blocking glass upper limb border;
(6) determine the image scanning region of located lateral scanning: using two longitudinal edge place straight lines of car plate as the left and right border in image scanning region, using in car plate top and move the line segment of 3～4 car plate width as the lower boundary in image scanning region, from top to bottom, scan from left to right described image scanning region, adding up the white pixel of every a line counts out, while extremely running into white pixel peak value for the first time, below this white pixel peak region, select a line as the coboundary in image scanning region, obtain thus whole image scanning region;
(7) definite image scanning region in scanning (6), adds up every row white pixel;
(8) according to transverse edge of peak value location wind-blocking glass of white pixel in statistics in step (7);
(9) repeating step (7), to (8), is located another transverse edge of wind-blocking glass;
(10) according to the wind-blocking glass lower edge border and the wind-blocking glass upper limb border that obtain in the transverse edge data of wind-blocking glass and step (5), obtain wind-blocking glass edge coordinate, wind-blocking glass region is cut apart in location.
In technique scheme, pretreatment operation described in step (1) also comprises carries out gray scale stretching, level and smooth, sharpening or medium filtering operation to image.
Preferred technical scheme, in step (6), the coboundary in image scanning region is positioned at white pixel peak region below, with the distance of white pixel peak region be 5～10 row.Owing to doing image pre-service before scanning, in image, large stretch of white space white noise of rear view of vehicle seldom, only have tailstock edge to assemble a lot of white pixel, it is white point pixel peak region, so in car plate two longitudinal boundaries from up to down during scan image, the white pixel number that every row counts at the beginning should be all fewer, and while being scanned up to that a few row in tailstock region, every row white point number there will be hop, be white pixel peak value, illustrate that coarse localization has arrived the tailstock, the coboundary that tailstock below one segment distance of take is like this rescan, just can avoid the more tailstock region of white point.
Wherein, the car plate rim detection of described step (2) refers to by mathematical computations and identifies the set that surrounding pixel gray-scale value has the pixel of step variation or roof variation.Edge is extensively present between object and background, between object and object, between picture element and primitive, be the key character that image is cut apart institute's reference.Car plate rim detection is prior art, and for example, Chinese invention patent application CN101408942A discloses the license plate locating method under a kind of complex background, comprises the steps: to carry out difference with the gray average in fixed size region, extracts car body; Car body image is carried out to the pre-service such as gray processing, gradient information enhancing, binaryzation; Pre-service aftercarriage image is carried out to mathematical morphological operation, in conjunction with license plate structure feature preliminary screening, obtain some candidate's license plate areas; The candidate's license plate area coloured image obtaining is converted into HSV color space image; Utilize BP neural network to carry out color identification to HSV color space candidate's license plate image, by calculating a certain pixel adjacent area average color distance, carry out colour edging detection; Edge point carries out edge color pair judgement, rejects false car plate edge; In conjunction with car plate textural characteristics, car plate is finally located.This method from the very detailed fractionation of the angle of mathematical model the pinpoint all processes of car plate.
The present invention has adopted Sobel operator when detecting car plate edge, Sobel operator can be upper and lower according to pixel, left and right neighbor pixel intensity-weighted Cha edge reaches this phenomenon of extreme value and detects edge, target area, and its horizontal direction operator Sx and vertical direction operator Sy can be expressed as:
,
Above shown in two convolution kernels formed Sobel boundary operator, each pixel in pretreated vehicle image will be done convolution with these two cores, Sx is maximum to horizontal edge response, Sy is maximum to vertical edge response.The maximal value of two convolution is as the output on this point, and the result calculating is like this exactly a breadths edge magnitude image.In addition, Sobel operator has smoothing effect for the noise in image, comparatively accurate edge directional information can be provided, compare with other operators (as Robert operator, Prewitt operator) for rim detection, more be applicable to the demand of the present invention to car plate rim detection.Binaryzation operation described in step (2) can be eliminated road surface and the interference of car body to car plate position to a great extent.The present invention adopts the complete threshold binarization algorithm of optimization, and this algorithm has good adaptivity under complex background.If the tonal range of a width vehicle image f (x, y) is (Z
1, Z
k), Z is gray-scale value, K is the numerical value number of the expression gray-scale value of appearance; If the threshold value T after optimizing
rz
1to Z
kbetween a number, this threshold value can obtain by alternative manner, R is iterations, specific algorithm is described below:
[a] obtains the minimum gradation value Z in image
minwith maximum gradation value Z
max, establish initial threshold value and be
。
[b] is according to threshold value T
rimage is divided into target and background two parts, and calculates two-part average gray value Z
aand Z
b.
Wherein, z (i, j) is the gray-scale value of image mid point (i, j), and N (i, j) is that gray-scale value is the number of pixels of z (i, j).
[c] calculates new threshold value
[d] is if T
r=T
r+1, T
rbe the threshold value after optimization, finish algorithm; Otherwise R ← R+1, forwards step [b] to.
By this threshold application, in the binaryzation operation of pending image, will there is phenomenon of rupture in car plate edge after binary conversion treatment.
In described step (6), by scanning vehicle image, reject the interference region (grid that for example dispel the heat, lane line, tailstock edge) on image, extract the target area of follow-up located lateral operation scanning, described method is:
[a] extends step (4) car plate longitudinal boundary, obtains longitudinal left and right border, target area.
[b] be 3～4 of translation car plate coboundaries car plate width distance upwards, as the horizontal lower boundary in target area, to reject the interference regions such as heat radiation grid.
[c] the determined scope of step [a] [b] from top to bottom, scan image from left to right, with array statistics[] add up the white pixel number of each scan line, while running into for the first time white pixel peak value, exit, specific algorithm is as follows:
[c-1] for (int h=0; H < picture altitude; H++)
[c-2] for (int w=scanning left margin; W <=scans right margin; W++)
[c-3] If (this point is white point) statistics[h] ++;
[c-4] If (statistics[h] >=critical value)
[c-5] preserves current line index;
[c-6] break；
[c-6] }
[c-7] }
Wherein, critical value is one and is slightly less than the numerical value of white pixel peak value for the first time, can obtain by test.This test mode is prior art, be explained as follows, in the image after binaryzation, except object edge place there will be white pixel peak value, other local white pixel numbers are all less and change and tend to be steady in scanning, critical value be one can differentiate between images in the white pixel number value of general position and object edge position, the method of getting critical value by test is very general in digital image processing techniques, for example, can be in the longitudinal boundary of left and right complete scan image from up to down, add up the white pixel number of every row, then line number is got to weighted mean value A
1, larger the closer to the row weight at tailstock edge, because the peak value that will find is just near the tailstock.Or in the longitudinal boundary of left and right, only scan the first half of vehicle image, line number is got to weighted mean value A
2.If think, a nearlyer step is accurate, can be to A
1with A
2be weighted on average, weight is A again
2>A
1.
[d] obtains the peak value index of preserving in [c], suitably reduces △ h, obtains a new height, makes it cross tailstock edge, is the transversely border of target area.
The region that four borders that obtain in step [a], [b], [d] are made is as the scanning area of follow-up wind-blocking glass located lateral.
The scan statistics operation of described step (7) (8) refers to, in the definite target area of step (6), image is carried out to rescan, adds up the white pixel number of each scan line, detects laterally upper and lower two edges of wind-blocking glass.Described scan statistics algorithm is:
[a] for (int h=scanning coboundary; H <=scans lower boundary; H++)
[b] for (int w=scanning left margin; W <=scans right margin; W++)
[c] If (this point is white point) statistics[h] ++;
[d] obtains peaked index Index in array statistics, preserves Index value;
[e] is made as (0,0,0) by the capable upper and lower 10 row rgb values of Index in image;
[f] if (finding wind-blocking glass coboundary & & to find wind-blocking glass lower boundary)
[g] finishes algorithm;
[h] else forwards step [a] to, continues;
Wherein, carrying out step [e] is mainly to consider that on wind-blocking glass, lower limb is not two real straight lines, but the tight continuous white pixel of several rows, may cause several provisional capitals that white pixel is maximum to be gathered in the situation of same edge, increased the difficulty of quick taking-up two edges, so its upper and lower 10 row pixels are all set to black when each preservation scan line index, guarantee that next time, the row of location there will not be near last minor peaks row.
Because technique scheme is used, the present invention compared with prior art has following advantages:
1. the present invention is by the relation in research license plate area and wind-blocking glass region, utilize car plate location technology first to orient license plate area, obtain accordingly image scanning region, again by the scanning to this region, utilize the edge in the method detection wind-blocking glass region of statistics specific pixel, can reach the accurate location in vehicle wind-blocking glass region; And then make traffic control personnel to area-of-interest see better, more accurate, improved their work efficiency, thereby greatly improved the comprehensive and real-time of whole intelligent transportation system.
2. the present invention has optimized car plate edge detection method, by the complete threshold binarization algorithm of optimizing, extracts car plate edge, makes result be more suitable for the demand that edge of the present invention extracts.
3. experiment showed, that the method that the vehicle wind-blocking glass location based on car plate location technology and image scanning statistics of the present invention is cut apart is feasible, and there is higher precision and execution efficiency.
4. the present invention is when wind-blocking glass and car plate location, adopted the statistical method, alternative manner and the simple matrix computations method that have compared with strongly-adaptive, avoided complicated loaded down with trivial details mathematical formulae and calculated, thereby reduced calculated amount, improve operation efficiency, met the requirement of real-time.
5. the present invention has not only considered the grey scale change of vehicle image, has also taken into account the variation of vehicle wind-blocking glass overall profile structure, thereby has solved the defect of general localization method precision deficiency, makes the more approaching and actual conditions of segmentation result.
Accompanying drawing explanation
Fig. 1 is the overall framework figure of embodiment of the present invention method.
Fig. 2 is the vehicle original image exemplary plot gathering in embodiment.
Fig. 3 carries out to Fig. 2 the image obtaining after pre-service.
Fig. 4 positions the image after processing to the car plate in Fig. 3.
Fig. 5 is the wind-blocking glass longitudinal register analysis chart in embodiment.
Fig. 6 is wind-blocking glass longitudinal register result in embodiment.
Fig. 7 dwindles targeted scans regional analysis figure in embodiment.
Fig. 8 is wind-blocking glass transverse edge testing result in embodiment.
Fig. 9 is wind-blocking glass edge positioning result and analysis chart in embodiment.
Figure 10 is wind-blocking glass segmentation result figure in embodiment.
Embodiment
Below in conjunction with drawings and Examples, the invention will be further described:
Embodiment: referring to accompanying drawing 1, for a kind of vehicle wind-blocking glass location dividing method overall system frame diagram based on car plate location technology and image scanning statistics that the present invention proposes, data file (picture file) is jpg or the bmp picture that reaches traffic control department Recognized Standards.The method of the present embodiment specifically comprises the following steps:
[1] collection of image: the requirement according to traffic control department to vehicle image quality standard, by ccd video camera, in traffic block port, charge station, absorb the vehicle image under multiple different automobile types, different colours, different light rays environment, with jpg or bmp form, preserve, and by getImage module reading images, referring to accompanying drawing 2.The number of getting vehicle pictures in this example is 273, and the resolution of individual picture is 1360 * 1024, and other image informations of traffic system needs.
[2] pre-service of image: in Bitmap bitmap object, preprocessing process comprises the processes such as gray processing, gray scale stretching, image smoothing, medium filtering, sharpening by the package images reading.Said process is outside the pale of civilization all optional for the normal image of acquisition quality except gray scale, only, for the image or some the particular procedure effect that have defect, just implements other preprocessing process.Pretreated image is referring to accompanying drawing 3.
[3] car plate location
[3-1] obtains the interference that the vehicle image after the processing that car plate edge amplitude image obtains in [2] has been removed color and light, only comprises colouring intensity information.Obtain Sobel operator in the horizontal direction with vertical direction on operator as convolution kernel.
，
With from left to right, top-down order, each pixel in image and this two operators are done to convolution, the maximal value of two convolution, as the output of this point, after computing finishes, obtains a vehicle edge magnitude image.
[3-2] calculated threshold binary image adopt iterative algorithm described in technical scheme to calculate applicable threshold value of the present invention, adopt the complete threshold binarization algorithm of optimizing, vehicle edge magnitude image is carried out to binary conversion treatment, and image will be in car plate, the fracture of wind-blocking glass edge generation color.
The location at [3-3] car plate edge adopts pattern matrix traversal binary image, find car plate four edges edge, an edge of car plate detected at every turn, record car plate edge coordinate xU, xD, yL, yR, and on this position, add red boost line, until mark four edges, limit of car plate, as shown in Figure 4.Wherein xU (xUp) is car plate coboundary on directions X in bitmap object; XD (xDown) is the lower boundary of car plate on bitmap directions X; YL (yLeft) is the left margin of car plate in bitmap Y-direction; YR (yRight) is the right margin of car plate in bitmap Y-direction.
[4] longitudinal register of wind-blocking glass extends two edges of the car plate vertical direction of obtaining in [3] referring to accompanying drawing 5, can pass wind-blocking glass region, and according to reality investigation and daily experience, the downward ab width of cyclecar wind-blocking glass is about 3 times of car plate width.If think, straight line l, r are divided into three parts by wind-blocking glass regions perpendicular, and the upper lower limb of center section is that line segment hg, ef are isometric, and the two-part length of upper edge in left and right is about 2/3 of lower limb length, i.e. dh=2/3ae, gc=2/3fb.According to the car plate boundary coordinate xU, xD, yL, the yR that obtain in [3-3], calculate car plate length HD (HorizontalDistance)=yR-yL in the horizontal direction, car plate width VD (VerticalDistance)=xD-xU in vertical direction.Wind-blocking glass lower limb middle part line segment L2 is respectively expanded to a HD length to the left and right both direction of level, by on wind-blocking glass along middle part line segment L1 as both sides each extension length 2/3HD, record coordinate, and carry out mark with green line, wind-blocking glass longitudinal register result is referring to accompanying drawing 6.
[5] located lateral of wind-blocking glass
[5-1] dwindles the scanning area that this step of scope to be scanned is intended to dwindle location, follow-up wind-blocking glass edge.
Referring to accompanying drawing 7, the longitudinal two edges of the car plate of usining straight line l, r are as the left and right scanning boundary of target area.Reason is as follows:
(1) object on vehicle edge and vehicle both sides road comprises a large amount of White lnterfere pixels, and above-mentioned sweep limit can be got rid of these interfere informations to the full extent;
(2) scanning process statistics is the number of every row white pixel point, and on wind-blocking glass, the white pixel point of lower limb is the most continuous and concentrated within the scope of straight line l-r.
Referring to accompanying drawing 7, using in car plate horizontal direction top and move the line segment DC of 3-4 car plate width as the scanning lower boundary of target area.Reason is as follows:
In the horizontal direction, a class interfere information derives from the car plate and heat radiation gate region that contains a large amount of white informations.Heat radiation grid are usually located at directly over car plate, and suitable with car plate width, according to the car plate width VD calculating in technical scheme, along 3-4 VD distance of X-axis negative direction vertical translation, make its coboundary move to line segment CD position car plate.Through a large amount of tests, mobile this length can be got rid of the interfere informations such as heat radiation grid, car plate outside sweep limit.
Referring to accompanying drawing 7, according to the algorithm of describing in technical scheme steps (6), scan image, obtains scanning coboundary, target area AB first.Reason is as follows:
In the horizontal direction, another kind of interfere information comes from vehicle tail edge, because the noise above image mostly is discrete distribution, so when being scanned up to tailstock edge, a large amount of white pixel continuously of this region clustering can become first peak value in scanning result, will as scanning coboundary, can reject the interfere information of the tailstock and further part thereof herein.
[5-2] detects wind-blocking glass transverse edge referring to accompanying drawing 7, according to the algorithm of describing in step in technical scheme (7)-(9), in [5-1], in definite target area ABCD, image is scanned, add up line by line white pixel number, each index that all obtains white pixel peak row, with this row of green line mark, until detect the two edges of keeping out the wind in recessed region horizontal direction, record coboundary horizontal ordinate TM (TopMargin), lower limb horizontal ordinate IM (InferiorMargin).Wind-blocking glass transverse edge testing result is referring to accompanying drawing 8.
[6] wind-blocking glass region cuts apart referring to accompanying drawing 9, according to the coordinate at wind-blocking glass four edges that obtain in [4] [5], from original vehicle image, be partitioned into a rectangular area that comprises wind-blocking glass, calculating four apex coordinates of wind-blocking glass is E (TM, ULM), F (TM, URM), G (IM, DRM), H (IM, DLM).With red line, connect 4 of EFGH and be partitioned into wind-blocking glass region.
[7] wind-blocking glass Region Segmentation experimental result accompanying drawing 10 has been shown the trapezoidal wind-blocking glass region being partitioned into from original figure, can find that segmentation effect is accurately effectively.
Experiment showed, that position success rate of the present invention and segmentation precision all have better effects, collect 273 different automobile types, the algorithm of locating license plate of vehicle of image applications the present embodiment that comprises complete vehicle body and car plate are tested, position success rate has reached 90% left and right.For car plate, locate successful image applications wind-blocking glass of the present invention location dividing method and test, success ratio has reached 98% left and right, and the wind-blocking glass region contour being partitioned into is clear, more highlights scene in car, the more important thing is greatly reducing of processing time.Apply the present invention in intelligent transportation system, make the result of whole system more comprehensive, more realistic application demand, thus minimizing traffic control personnel's labor workload has improved work efficiency.
Claims (2)
1. the vehicle wind-blocking glass region in image is positioned to a method of cutting apart, it is characterized in that, based on car plate location technology and image scanning statistics, realize, comprise the following steps:
(1) obtain image and carry out pretreatment operation, described pretreatment operation comprises carries out gray processing to image;
(2) with Sobel operator, detect car plate edge, binary image;
(3) take described Sobel operator as pattern matrix coupling car plate matrix of edge, positioning licence plate;
(4) obtain car plate edge coordinate, extract coordinate information, and calculate and obtain length in the horizontal direction of car plate and the width in vertical direction;
(5) line segment of car plate horizontal direction is respectively expanded to a length to the left and right both direction of level, each expands a two-end-point place vertical line after length is wind-blocking glass lower edge border, the line segment of car plate horizontal direction is respectively expanded to 2/3 length to the left and right both direction of level, and each expands 2/3 two-end-point place vertical line after length is wind-blocking glass upper limb border; Described length is car plate length in the horizontal direction;
(6) determine the image scanning region of located lateral scanning: using two longitudinal edge place straight lines of car plate as the left side in image scanning region, right margin, using in car plate top and move the line segment of 3～4 car plates width in vertical direction as the lower boundary in image scanning region, from top to bottom, the image scanning region of coboundary is not yet determined in scanning from left to right, adding up the white pixel of every a line counts out, while extremely running into white pixel peak value for the first time, below being expert at this white pixel peak value, select a line as the coboundary in image scanning region, obtain thus whole image scanning region, wherein, the coboundary in image scanning region be positioned at white pixel peak value be expert at below, the distance of being expert at white pixel peak value is 5～10 row,
(7) definite image scanning region in scanning (6), adds up every row white pixel;
(8) according to transverse edge of peak value location wind-blocking glass of white pixel in statistics in step (7);
(9) repeating step (7), according to another transverse edge of the peak value location wind-blocking glass of white pixel in statistics in step (7);
(10) according to the wind-blocking glass lower edge border and the wind-blocking glass upper limb border that obtain in the transverse edge data of wind-blocking glass and step (5), obtain wind-blocking glass edge coordinate, wind-blocking glass region is cut apart in location.
2. according to claim 1 the vehicle wind-blocking glass region in image is positioned to the method for cutting apart, it is characterized in that: pretreatment operation described in step (1) also comprises carries out gray scale stretching, level and smooth, sharpening or medium filtering operation to image.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201210120685.4A CN102663760B (en) | 2012-04-23 | 2012-04-23 | Location and segmentation method for windshield area of vehicle in images |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201210120685.4A CN102663760B (en) | 2012-04-23 | 2012-04-23 | Location and segmentation method for windshield area of vehicle in images |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN102663760A CN102663760A (en) | 2012-09-12 |
| CN102663760B true CN102663760B (en) | 2014-08-20 |
Family
ID=46773238
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201210120685.4A Expired - Fee Related CN102663760B (en) | 2012-04-23 | 2012-04-23 | Location and segmentation method for windshield area of vehicle in images |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN102663760B (en) |
Families Citing this family (15)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102999749B (en) * | 2012-12-21 | 2015-10-28 | 广东万安科技股份有限公司 | Based on the securing band violation event intelligent detecting method of Face datection |
| CN103207988B (en) * | 2013-03-06 | 2016-02-10 | 大唐移动通信设备有限公司 | A kind of image-recognizing method and device |
| CN103324935B (en) * | 2013-06-27 | 2016-12-28 | 中山大学 | Vehicle is carried out the method and system of location and region segmentation by a kind of image |
| CN103440647B (en) * | 2013-08-20 | 2016-12-07 | 中山大学 | The detection method of a kind of vehicle sun visor state and system |
| CN105260701B (en) * | 2015-09-14 | 2019-01-29 | 中电海康集团有限公司 | A kind of front vehicles detection method suitable under complex scene |
| CN109960984A (en) * | 2017-12-25 | 2019-07-02 | 大连楼兰科技股份有限公司 | Vehicle checking method based on contrast and significance analysis |
| CN108428246B (en) * | 2018-03-09 | 2020-10-09 | 安徽农业大学 | Method for acquiring area of front windshield for automobile defrosting test in low-temperature cabin environment |
| CN109063572A (en) * | 2018-07-04 | 2018-12-21 | 南京信息工程大学 | It is a kind of based on multiple dimensioned and multireel lamination Fusion Features fingerprint activity test methods |
| CN109165600B (en) * | 2018-08-27 | 2021-11-26 | 浙江大丰实业股份有限公司 | Intelligent search platform for stage performance personnel |
| CN109255318A (en) * | 2018-08-31 | 2019-01-22 | 南京信息工程大学 | Based on multiple dimensioned and multireel lamination Fusion Features fingerprint activity test methods |
| CN109886271B (en) * | 2019-01-22 | 2021-01-26 | 浙江大学 | Image accurate segmentation method integrating deep learning network and improving edge detection |
| CN109740611A (en) * | 2019-01-25 | 2019-05-10 | 中电健康云科技有限公司 | Tongue image analysis method and device |
| CN110473221B (en) * | 2019-08-20 | 2020-04-24 | 吕若丹 | Automatic target object scanning system and method |
| CN111724561A (en) * | 2020-05-19 | 2020-09-29 | 中国安全生产科学研究院 | Sulfide ore spontaneous combustion monitoring method and system |
| CN114038208B (en) * | 2021-11-04 | 2022-08-26 | 重庆畅停智慧停车服务有限公司 | Roadside parking lot fee evasion management method |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US7139410B2 (en) * | 2001-10-30 | 2006-11-21 | Denso Corporation | Apparatus for protecting occupant in vehicle |
| CN101504717A (en) * | 2008-07-28 | 2009-08-12 | 上海高德威智能交通系统有限公司 | Characteristic area positioning method, car body color depth and color recognition method |
| CN102169583A (en) * | 2011-05-13 | 2011-08-31 | 北方工业大学 | Vehicle shielding detection and segmentation method based on vehicle window positioning |
-
2012
- 2012-04-23 CN CN201210120685.4A patent/CN102663760B/en not_active Expired - Fee Related
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US7139410B2 (en) * | 2001-10-30 | 2006-11-21 | Denso Corporation | Apparatus for protecting occupant in vehicle |
| CN101504717A (en) * | 2008-07-28 | 2009-08-12 | 上海高德威智能交通系统有限公司 | Characteristic area positioning method, car body color depth and color recognition method |
| CN102169583A (en) * | 2011-05-13 | 2011-08-31 | 北方工业大学 | Vehicle shielding detection and segmentation method based on vehicle window positioning |
Non-Patent Citations (4)
| Title |
|---|
| A Study on Detection of Vehicles" Positions Using HSV Color Transformed Template Images;Hiroaki FUKUI等;《2001.VTC 2001 Fall.IEEE VTS 54th Vehicular Techonlogy Conference》;20111011;第2卷;718-722 * |
| Hiroaki FUKUI等.A Study on Detection of Vehicles" Positions Using HSV Color Transformed Template Images.《2001.VTC 2001 Fall.IEEE VTS 54th Vehicular Techonlogy Conference》.2011,第2卷718-722. |
| 一种车窗定位算法的设计和实现;张轶等;《西安航空技术高等专科学校学报》;20060120;第24卷(第1期);12-14 * |
| 张轶等.一种车窗定位算法的设计和实现.《西安航空技术高等专科学校学报》.2006,第24卷(第1期),12-14. |
Also Published As
| Publication number | Publication date |
|---|---|
| CN102663760A (en) | 2012-09-12 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN102663760B (en) | Location and segmentation method for windshield area of vehicle in images | |
| CN106709436B (en) | Track traffic panoramic monitoring-oriented cross-camera suspicious pedestrian target tracking system | |
| CN103927526B (en) | Vehicle detecting method based on Gauss difference multi-scale edge fusion | |
| KR101403876B1 (en) | Method and Apparatus for Vehicle License Plate Recognition | |
| CN101334836B (en) | License plate positioning method incorporating color, size and texture characteristic | |
| CN105488454B (en) | Front vehicles detection and ranging based on monocular vision | |
| CN102375982B (en) | Multi-character characteristic fused license plate positioning method | |
| CN102509098B (en) | Fisheye image vehicle identification method | |
| CN104021574B (en) | Pavement disease automatic identifying method | |
| CN110210451B (en) | Zebra crossing detection method | |
| CN105488492B (en) | A kind of color image preprocess method, roads recognition method and relevant apparatus | |
| CN104036262B (en) | A kind of method and system of LPR car plates screening identification | |
| CN103903018A (en) | Method and system for positioning license plate in complex scene | |
| CN106650553A (en) | License plate recognition method and system | |
| CN110414355A (en) | The right bit sky parking stall of view-based access control model and parking stall line detecting method during parking | |
| Alonso et al. | Robust vehicle detection through multidimensional classification for on board video based systems | |
| CN102842037A (en) | Method for removing vehicle shadow based on multi-feature fusion | |
| CN102073852B (en) | Multiple vehicle segmentation method based on optimum threshold values and random labeling method for multiple vehicles | |
| CN103116751A (en) | Automatic license plate character recognition method | |
| CN108052904B (en) | Method and device for acquiring lane line | |
| CN106887004A (en) | A kind of method for detecting lane lines based on Block- matching | |
| CN105005766A (en) | Vehicle body color identification method | |
| CN104700072A (en) | Lane line historical frame recognition method | |
| CN103530608A (en) | Vehicle type judgment method and vehicle type judgment device | |
| CN104809433A (en) | Zebra stripe detection method based on maximum stable region and random sampling |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| C14 | Grant of patent or utility model | ||
| GR01 | Patent grant | ||
| CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20140820 Termination date: 20180423 |
|
| CF01 | Termination of patent right due to non-payment of annual fee |